On Wed, 1 Sep 2010, Mohan V wrote: > Hello All, > > According to the USB 2.0 spec, "bInterval" value for the High Speed Bulk > OUT endpoint must specify the maximum NAK rate of the endpoint. (pg. > 271). But on page 273 of the specs, its written > "For high-speed bulk and control OUT endpoints, the bInterval field is > only used for compliance purposes; the host controller is not required > to change its behavior based on the value in this field." > > I wanted to know, for High speed Bulk OUT endpoints, what should be the > value of bInterval. It _should_ be the value defined in the spec, i.e., the maximum NAK rate. > If it is set to 0, it indicates that the endpoint never NAKs. So, does > this mean that device can never NAK? It means that _endpoint_ will never NAK (assuming the host behaves correctly). The device can still NAK on other endpoints. > Consider the following case: > The host is sending data, device sends an ACK for first chunk; > The host will send the next chunk; if for some reason, the device is not > able to receive data, will the host retry if bInterval = 0? Yes, it will. All the host controllers I know about ignore the bInterval value for bulk endpoints. However by specifying bInterval = 0 and returning ACK for the first chunk, the device has promised that it _will_ be able to receive the second chunk. So this case should never occur. > I am asking this by considering the USB android gadget composite driver, > gadget/f_mass_storage.c where the bInterval is set to 1 for high speed > bulk out endpoint for mass storage interface. I believe that value was chosen simply as a default to be safe. It doesn't really mean anything, since it would have to depend on the device controller hardware. Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
